Statistical Combat Analysis

[crzyrndm]

I want to spend more time reading through this thread but I have a couple quick questions:

• 40 attack/80 defence appears to be an outlier

Does this mean that one would need 80 defense or greater to achieve a proper defense against 40 attack?

It was a note that that particular experimental data point didn't follow the normal pattern, which was quite noticable on the graph (the purple line kinks downwards for that point). There is no point for "proper defence" as such. Increasing defence up into the hundreds continues to decrease the hit chance (depending on the attack value).

[gdwitt]

This is an important study.
The obvious question is why do we even have defense values if the only thing important is the differense between attack and defence.
Why not keep the older statof 1 melee value?
Combine shield and defense and attack?
If we're stuck with the current system, can we use a ratio defense:attack to use in balancing units?

[Destin Faroda]

The obvious question is why do we even have defense values if the only thing important is the differense between attack and defence.

Defense does work properly if you use reasonable unit stats and not the mess CA has created.

If we're stuck with the current system, can we use a ratio defense:attack to use in balancing units?

In Shogun II, I use attack values between 3 (pikes) and 7 (odachi). Melee defense is between 4 (pikes) and 7 (swords). Modifiers like unit quality, experience and light armour can increase the defense value. The armour system from RealCombat 2 works very well with Warscape games. Usually, defence should be several points higher than attack values. In combination with 2 hitpoints per unit, JaM's morale system and slower movement speed, Warscape combat starts to get enjoyable.

[gdwitt]

Defense does work properly if you use reasonable unit stats and not the mess CA has created.

I don't think you understand.
What if you eliminated 1 variable and just called the other one "melee"?
What if defense is really only a poor way to add attack to a unit?

I just attacked a force of low-tier sword units with the best 1st turn Athenian hoplites having hi armor and defense. We each had 2 slingers.
Athenian hoplites 14 attack, 45 defense, 50 armor, bronze shield, phalanx ability, morale at 45.
Their units were non-ap and 18 at 18 defense and had a numerical advantage of 2 extra units with less morale and much lower armor/shield values.
I did my best to eliminate being attacked in the back.
I was routed with losses about 40% more than their army.

Hypothesis: Defense does nothing but add more attack. The other values add to armor which only negates damage.
I don't know how to test this, but would guess a progression of differences between two with each having an attack value like differences of 10, 20, 30, 40
Here's a valuable test I'd like to help with if possible:
1)Compare a unit with 0 attack 10 defense with 0 attack 20 defense. Follow the progression.
2)Compare now a progression of defense to attack with the same ration. 0at/10def vs 10/0 def., 10/20 vs20/10, 20/30 vs 30/20.... etc.
3)Compare this progression with same at/def difference: of 0at/10def vs10/0; 10/10 vs 20/0; 20/10 vs. 30/0, 30/10 vs. 40/0.
If the winning values are the same, this one might prove that defense has no kill value because the attack unit has 0 defense.
Now what if high defense values only worked with a small attack value as we see with the hoplites.
4) Compare progression with rising at/def: 0at/10def vs 10/0 as control; 10/10 vs 10/0; 10/20 vs 10/0; 10/30 vs. 10/0, 10/40 vs. 10/0, 10/50 vs 10/0

Do you get the idea?
Which one of these would be worth testing?

[crzyrndm]

What if defense is really only a poor way to add attack to a unit?

Melee Defence does the opposite of Melee Attack, so I don't quite understand how this question comes about.

This is proven by the increasing duration of combat with increasing defence, and decreasing duration of combat with increasing attack.

[gdwitt]

I know how to edit pack files, but how do I choose the units in combat in custom battles without choosing a commander?
What other settings can you advise that will help me test?

Does melee defense actually cause damage or kills or is it just useful in preventing a successful collision event?
If a unit had 0 attack, will it be able to cause any damage?

[Lemonater47]

The part about shields is now out of date. Out of date since patch 5.

Sheilds now have a chance of blocking projectiles. Meaning that no damage is dealt when the shield is hit by a projectiles. That's if the dice roll says so. When I last looked roman shields had a 55% chance of blocking projectiles. Other shields have a lower chance.
So read the patch 5 notes balancing section and its the first battle balance change on the list.

[crzyrndm]

Right, thanks for reminding me. I keep meaning to update the mechanics part but never seem to get to it

[gdwitt]

That's a great mod.
I really want to try a study of sort since I'm new to the game and want to start modding.
The key I need is how to balance the spear versus sword warfare. It seems that the hoplite spear (lo attack/hi defense) is too weak against cheaper sword units even when I rule out envelopment around the sides.
I was playing on hard difficulty in battle.

What kind of study would you suggest?
Here are some examples:
1)Compare a unit with 0 attack 10 defense with 0 attack 20 defense. Follow the progression.
2)Compare now a progression of defense to attack with the same ration. 0at/10def vs 10/0 def., 10/20 vs20/10, 20/30 vs 30/20.... etc.
3)Compare this progression with same at/def difference: of 0at/10def vs10/0; 10/10 vs 20/0; 20/10 vs. 30/0, 30/10 vs. 40/0.
If the winning values are the same, this one might prove that defense has no kill value because the attack unit has 0 defense.
Now what if high defense values only worked with a small attack value as we see with the hoplites.
4) Compare progression with rising at/def: 0at/10def vs 10/0 as control; 10/10 vs 10/0; 10/20 vs 10/0; 10/30 vs. 10/0, 10/40 vs. 10/0,

I hope you get the idea.

[Roach]

Are the 15% - 85% hit chance limits the post patch 9 numbers?

[Magnar]

to get a real idea of how these stats effect each other, max hit chance should be set to 100 and minimum 0. .. did u do that? or is it based on vanilla values?

[crzyrndm]

to get a real idea of how these stats effect each other, max hit chance should be set to 100 and minimum 0. .. did u do that? or is it based on vanilla values?

This was done when vanilla had used 95 and 20. I don't see how giving another 5% range that never gets used at the top end will show anything, nevermind the the lower cap that is only reached with 0 defence.

[phaionix]

So I guess it's known that bonus vs infantry adds to both melee attack and melee damage against infantry, but do we know whether it's ap damage or just normal damage?

[crzyrndm]

From my testing the bonus vs. only added damage (?), and it's normal damage, not ap

[phaionix]

Ah okay. I think I was confused by some conflicting info from this thread or I might just have misread something. Thanks! Keep up the good work!

[Tsubasa]

@crzyrndm
Some little questions please!

Are the values of Melee Attack and Melee Defence always fixed Values for calculation and not a random number in between there value-borders?
And how it is looking with Weapon Damage and Armour? Actually i believe just Armour is a random number. Is this true?

[crzyrndm]

I do not know how melee attack/defence calculations work exactly.
Jack lusted said that armour was the randomised value of the damage/armour set.

What is happening doesn't exactly matter though.